Browse Prior Art Database

System and Method for Summarizing Symptoms for IT incidents

IP.com Disclosure Number: IPCOM000250112D
Publication Date: 2017-Jun-01
Document File: 3 page(s) / 727K

Publishing Venue

The IP.com Prior Art Database

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 55% of the total text.

System and Method for Summarizing Symptoms for IT incidents

Abstract:

In the domain of Information Technology (IT) incident management, each IT

incident has description of the reported IT issues, resolution, and root cause

analysis, which are written in free text. Incident summarization converts

unstructured text into structured one, which is a critical step for incident

management. However, existing summarization techniques were developed to

analyze well-written articles like news, journals, research papers, etc. To help

practitioners to diagnose and prevent IT incidents, this disclosure proposes a

common approach to summarize historical IT incidents to extract meaningful

keywords that precisely capture each incident.

Body:

In the domain of Information Technology (IT) incident management, each IT

incident has description of the reported IT issues, resolution, and root cause

analysis, which are written in free text. Through summarization of an IT incident,

unstructured text can be converted into structured one, from which we can

detect common IT issues, problematic areas, and proven practices to resolve

incidents. Thus, incident summarization is a critical step for incident

management. However, existing summarization techniques were developed to

analyze well-written articles like news, journals, research papers, etc. For IT

incidents, the text is a mixture of machine generated code and plain English

text. Particularly, the text contains lots of technical terms, ambiguous terms,

acronyms etc. The existing text mining techniques are ineffective in this domain.

To address the above-mentioned challenges, this disclosure develops an

approach to explore a collection of Information Technology (IT) incidents

through underlying topics or themes that run through the collection. It contains

three major steps, as shown in Figure 1.

Figure 1. Major flow.

Step 1: Generate candidate topics using Probabilistic Topic Model

At this step, we apply Probabilistic Topic Model to generate candidate topics

from the unstructured text. In particular, we propose to use Latent Dirichlet

Allocation (LDA), one typical Probabilistic Topic Models, to gen...